摘要
采用Daubechies4小波对图像进行三层Mallat塔式分解,取每个分解层次上的每个子带图像的能量,加入在低频子带上提取的灰度共生矩阵统计量,来形成最终的特征向量以提高检索精度。
Texture likelihood research is an important part of the content-based image retrieval research. Adoptes wavelet Daubechies4 to conduct a three level Mallat pyramid decomposition, takes the energy of each subband image in each decomposition level and adds in the gray level co-occurrence matrix statistics extracted in the low-frequency subband to become the final feature vector in order to increase the retrieval accuracy.
出处
《现代计算机》
2007年第10期58-59,共2页
Modern Computer
关键词
纹理相似性
Mallat塔式分解
子带图像
灰度共生矩阵
Texture Likelihood
Mallat Pyramid Decomposition
Subband Image
Gray Level Co-occurrence Matrix